4 Comparing Party Systems PETER MAIR
The comparison of party systems has always formed an essential component of the more generalized comparison of democratic political systems. On the one hand, by noting the number of parties in competition in any given polity, and by taking at least some account of the manner in which these parties interact with one another, it has proved possible to gain a reasonably valuable insight into the ways in which these polities differ from one another. On the other hand, following a more normative imperative, it has often proved tempting to trace the source of problems of the legitimacy and stability of democratic regimes back to the character of their party systems. For both reasons, an understanding of the nature and character of a country’s party system has always been accorded priority in cross-national comparative analysis, even if the criteria by which party systems are compared have often been subject to debate. This chapter begins with a review of the principal existing approaches to the classification of party systems, pointing to both their limits and possibilities when applied within comparative analysis. It then goes on, in the second section of the chapter, to underline the importance of understanding the structure of competition in any given party system, since in many ways the very notion of a party system is centred on the assumption that there exists a stable structure of competition. As is shown here, structures of competition can be seen to be either closed (and predictable) or open (and unpredictable), depending on the patterns of alternation in government, the degree of innovation or persistence in processes of government formation, and the range of parties gaining access to government. Given the overall concern of this volume, the emphasis in the third section is on the need to distinguish between processes of electoral change on the one hand, and changes in party systems and the structures of competition on the other, a distinction which also allows us to conceive of situations in which electoral change is the consequence rather than the cause of party system change.
C O M PA R I N G PA R T Y S Y S T E M S
Approaches to the classification of party systems: a review The most traditional and most widely accepted criterion for classifying party systems is also the most simple: the number of parties in competition. Moreover, the most conventional distinction involved here also proves appealingly straightforward: that between a two-party system on the one hand, and a multiparty system (that is, more than two parties) on the other (see Duverger 1954). When first promoted as the principal means of distinguishing between party systems, this mode of categorization was seen as tapping into a more fundamental distinction between more or less stable and consensual democracies, which were those normally associated with the two-party type, as opposed to more or less unstable and conflictual democracies, which were those associated with the multiparty type. Thus, and particularly in early applications, two-party systems, which were typically characteristic of the United Kingdom and the United States, and which invariably involved single-party government, were assumed to enhance accountability, alternation in government, and moderate, centre-seeking competition. Multiparty systems, on the other hand, which usually required coalition administrations, and which were typically characteristic of countries such as France or Italy, prevented voters from gaining a direct voice in the formation of governments, did not necessarily facilitate alternation in government, and sometimes favoured extremist, ideological confrontations between narrowly based political parties. And although this simple association of party system types and political stability and efficacy was later challenged by research into the experiences of some of the smaller European democracies in particular, which boasted both a multiplicity of parties and a strong commitment to consensual government (e.g., Daalder 1983), the traditional categorization of two-party versus multiparty has nevertheless continued to command a great deal of support within the literature on comparative politics.1 This simple distinction is, of course, far from being the only possible approach, and since Duverger a number of attempts have been made to develop more sensitive and discriminating criteria (see Table 4.1). In the conclusion to his classic Oppositions volume, for example, Robert Dahl (1966) sought to move away from an almost exclusive concern with simply the numbers of parties, and built an alternative classification based around the competitive strategy adopted by the opposing parties, distinguishing between competitive, co-operative, and coalescent strategies, and distinguishing further between opposition in the electoral arena and opposition in the parliamentary arena (see also Laver 1989). This led Dahl to elaborate a four-fold typology, distinguishing between strictly competitive systems, co-operative–competitive systems, coalescent–competitive systems, and strictly coalescent systems. Shortly after this, in what proved subsequently a very influential study, Jean Blondel (1968) developed a typology which took account not only of the numbers of parties in competition, but also
89
90
C O M PA R I N G D E M O C R A C I E S
TABLE 4.1
2
Types of Party Systems
Author Duverger (1954)
Principal criteria for classification Numbers of parties
Dahl (1966)
Competitiveness of opposition
Blondel (1968)
Numbers of parties Relative size of parties
Rokkan (1968)
Numbers of parties Likelihood of single-party majorities Distribution of minority party strengths
Sartori (1976)
Numbers of parties Ideological distance
Principal types of party system identified Two-party systems Multiparty systems Strictly competitive Co-operative–competitive Coalescent–competitive Strictly coalescent Two-party systems Two-and-a-half-party systems Multiparty systems with one dominant party Multiparty systems without dominant party The British-German “1 vs. 1 + 1” system The Scandinavian “1 vs. 3–4” system Even multiparty systems: “1 vs. 1 vs. 1 + 2–3” Two-party systems Moderate pluralism Polarized pluralism Predominant-party systems
their relative size (and, in a later refinement, their “place on the ideological spectrum”), distinguishing four types: two-party systems, two-and-a-halfparty systems, multiparty systems with a dominant party and multiparty systems without a dominant party. In practice, however, this new approach did little more than improve the traditional two-party versus multiparty distinction by disaggregating the otherwise overloaded multiparty category. Stein Rokkan’s (1968) contemporaneous attempt to classify the party systems of the smaller European democracies also did little more than disaggregate the multiparty category, in this case by taking account of the likelihood of single party majorities (akin to Blondel’s dominant party) and the degree to which there was a fragmentation of minority party strengths. Using these criteria, Rokkan developed a three-fold distinction involving a “British-German” type system, in which the system was dominated by the competition between two major parties, with a third, minor party also in contention; a “Scandinavian”-type system in which one big party regularly confronted a more or less formalized alliance between three or four smaller parties; and an “even” multiparty system in which competition was dominated by three or more parties of equivalent size. Counting the numbers of parties—or what he referred to as the “format” of the party system—is also an essential component of the more comprehensive typology which was later developed by Sartori (1976: 117–323).2 But although Sartori’s approach underlined the relevance of party numbers, it also went much beyond this by including as a second
C O M PA R I N G PA R T Y S Y S T E M S
principal criterion the ideological distance that separated the parties in the system. In fact, Sartori’s typology, which was explicitly concerned with the interactions between the parties in any given system—what he refers to as the “mechanics” of the system—and which was therefore explicitly concerned with differential patterns of competition, drew on the combination of these two criteria. Party systems could therefore be classified according to the number of parties in the system, in which there was a distinction between formats with two parties, those with up to five parties (limited pluralism) and those with six parties or more (extreme pluralism); and according to the ideological distance separating the extreme parties in the system, which would either be small (moderate) or large (polarized). The two criteria were not wholly independent, however, in that, as Sartori also argued, the format of the system, that is the number of parties, contained mechanical predispositions (that is, it could affect the degree of polarization), such that extreme pluralism could lead to polarization. The combination of both criteria then yielded three principal types of party system—two-party systems, characterized by an evidently limited format and a small ideological distance (for example, the United Kingdom); moderate pluralism, characterized by limited pluralism and a relatively small ideological distance (such as Denmark); and, which was the most important for the typology, polarized pluralism, characterized by extreme pluralism and a large ideological distance (examples being Italy in the 1960s and 1970s, and Chile prior to the 1973 coup). In addition, Sartori also allowed for the existence of a “predominant-party system,” a system in which one particular party, such as, most notably, Congress in India or the Liberal Democrats in Japan prior to the 1990s, consistently (that is, over at least four legislatures) won a majority of parliamentary seats.3 There are a number of reasons why Sartori’s typology can be regarded as the most important of those briefly reviewed here. In the first place, it is the most comprehensive of all the available typologies, both in terms of the care with which it is developed, as well as in terms of the way in which it is applied to empirical cases. Second, notwithstanding the continued appeal of the simple two-party/multiparty distinction, Sartori’s typology has subsequently been employed in a variety of sophisticated national and crossnational studies, yielding a degree of insight into the functioning of party systems which is incomparably better than that developed by any of the alternative typologies (e.g., Bartolini 1984; Bille 1990). Third, it underlines the influence exerted by systemic properties, and by the party system, on electoral behavior and electoral outcomes. More than any of the other typologies, it therefore allows the party system to operate as an independent variable, constraining or even directing electoral preferences. This last aspect is particularly important in this context, and I will return to it at a later stage. Finally, as noted, it is a typology that is explicitly concerned with patterns of competition and with the interactions between parties, and in this sense it is much more directly concerned with the functioning of the party system itself. Indeed, Sartori’s definition of a party system,
91
92
C O M PA R I N G D E M O C R A C I E S
2
formulated in his 1976 volume, still stands as the most precise and complete definition within the literature: Parties make for a “system”, then, only when they are parts (in the plural); and a party system is precisely the system of interactions resulting from inter-party competition. That is, the system in question bears on the relatedness of parties to each other, on how each party is a function (in the mathematical sense) of the other parties and reacts, competitively or otherwise, to the other parties. (Sartori 1976: 44)
At the same time, however, and some 25 years after the publication of Sartori’s seminal volume, questions can be raised regarding the continued utility and discriminating capacity of the typology, not least because of what is now a potential overcrowding in the moderate pluralism category and a virtual emptying of the alternative types. For example, and this criticism can also be levelled against the traditional Duverger classification, it is now relatively difficult to find clear-cut examples of the classic two-party system. The United States, which is often cited as an almost pure two-party model, can also be understood as a “four-party” system, in which a presidential two-party system co-exists with a separate congressional two-party system, or even as having 50 two-party systems, each functioning separately in each of the 50 states (e.g., Katz and Kolodny 1994). The United Kingdom, which was also always seen as the paramount case of a two-party system, recently, albeit temporarily, fulfilled Sartori’s conditions for a predominant party system, with the Conservative party consistently winning majorities in the 1979, 1983, 1987, and 1992 Westminster elections. Against this, of course, it might be argued that a number of newer democracies are beginning to approximate the two-party model, and that this could restore the relevance of the category as a whole. Since becoming full democracies in the 1970s, for example, Greece, Portugal, and Spain have all tended to drift increasingly towards a two-party format with alternating single-party governments. The Czech Republic may yet go in a similar direction, in that the two biggest Czech parties, the Civic Democrats and the Social Democrats, currently commanding some two-thirds of the seats in parliament, are together pushing for a reform of the electoral system which would effectively strengthen their own position and weaken that of their smaller opponents. Costa Rica, one of the most well established Latin American democracies, also reflects a two-party format, with the two biggest parties, the Christian-Social Unity party and the National Liberation party together winning more than 75% of the vote and some 90% of the seats in the 1998 Assembly elections. The purest two-party system is that found in Malta, one of the world’s smallest democracies, where the two main parties together poll almost exactly 100% of the vote. With this handful of exceptions, however, the two-party category is almost complete. At the other extreme, and particularly given the recent decline and/or eclipse of traditional communist parties, it has also become difficult to find
C O M PA R I N G PA R T Y S Y S T E M S
an unambiguous example of polarized pluralism. Sartori’s criteria for this latter system had been very carefully elaborated (1976: 131–73), and depended crucially on there being a “maximum spread of [ideological] opinion” (p. 135), bilateral oppositions (p. 134), and hence, necessarily so, on there being a relevant anti-system party, that is, a party which “undermines the legitimacy of the regime it opposes” (p. 133) at each end of the political spectrum. It follows from this that should either of these antisystem alternatives become irrelevant or disappear, there would then occur an attenuation of the spread of opinion and thus a reduction in the degree of polarization, so forcing the case out of the category. This is now certainly the case in France, for example, where the fading anti-system party of the left, the Communist party (PCF), was sufficiently legitimated to be admitted to government office as long ago as 1981; and, more recently, in Italy, where the Communist party (PCI) divided into the unequivocally moderate Democratic Left (DS) and the smaller, more radical, but certainly no longer anti-system alternative of the Communist Refoundation (RC). In addition, with the advent of the former neo-fascist National Alliance (AN) to office in the Berlusconi government of 1994, Italy can also be seen to have shed its anti-system alternative of the right. This is not to suggest that antisystem oppositions have everywhere ceased to exist; on the contrary, despite the eclipse of the traditional anti-system parties of the communist and fascist variety, a number of European party systems are now confronted with the rise of new parties, particularly on the right, which might well be seen as anti-system in orientation, such as the National Front in France or the Vlaams Blok in Belgium (Betz 1994; Ignazi 1992, 1994; Mudde 2000). But even if these parties do reflect an extreme of opinion on the right-wing side of the political spectrum, they tend not to be counter-balanced by an equivalent anti-system extreme on the left, and hence, by definition, the poles are no longer “two poles apart” (Sartori 1976: 135). In short, if two-party systems in a strict sense are thin on the ground, and if examples of polarized pluralism are also increasingly hard to find, then, perforce, most systems tend to crowd into the category of moderate pluralism,4 which clearly reduces the discriminating power of the typology.
Party systems and the structure of competition That said, Sartori’s approach remains particularly useful and important in that, unlike the alternative approaches, it helps to focus attention directly on what is perhaps the most important aspect of party systems, and on what distinguishes most clearly between different party systems: the structure of inter-party competition, and especially the competition for government. To be sure, it might be argued that this is in fact the core variable underlying each of the other established classifications of party systems. Duverger’s (1954) classic distinction between two-party systems and multiparty systems, for example, is obviously based on the numbers of parties in
93
94
C O M PA R I N G D E M O C R A C I E S
2
competition, but it can also be seen as differentiating systems in which two major parties compete with one another over the question of which will form a (single-party) government from those in which government usually involves some form of shifting coalition. In a similar vein, Rokkan’s (1968) distinction between the British-German, Scandinavian, and even the multiparty types, though ostensibly, and within his own terms, “purely numerical,” is really an attempt to tap into the different patterns of coalition formation. But while some notion of the competition for government may well have informed these earlier classifications, they certainly did not confront the issue directly. Indeed, among alternative approaches, it is really only Dahl’s (1966) distinctions that come anywhere close to addressing the question of government formation as a key defining feature, even though within his particular analysis this takes second place to the more central question involved in identifying differences in party strategies in different competitive arenas. Building from Sartori, then, how might we best understand those differential patterns in the competition for government that can be seen to define the character of party systems? What is important here is that we not only have a set of criteria which can distinguish between differential patterns of competition for government, but that we also have something that is sensitive to shifts over time. That is, we need criteria which can be used to categorize party systems in different polities, and which, at the same time, can allow us to see when party systems change. This involves not only placing competition for government at the heart of the definition of party systems, but also involves adopting a more dynamic perspective that can move away from the rather static categories that have tended to dominate the literature to date. Seen from this perspective, there are three factors that can be considered relevant. In the first place, there is the question of the prevailing mode of government alternation, since it is the competition between potential teams of governors which lies at the core of any party system. Second, there is the question of stability or consistency in the governing alternatives, and the extent to which competition for government takes on a familiar, and hence predictable character. Finally, there is the simple question of who governs, and the extent to which access to government is either open to a wide range of diverse parties or limited simply to a smaller subset. Let us now look at each of these factors in turn and at how they serve to shape the logic of any given party system.5
Alternation in government There are three conceivable patterns of alternation which need to be considered here. The first, and most obvious pattern might be termed wholesale alternation, in which incumbents are wholly displaced by a former opposition. In other words, all of the parties in government at time t are removed
C O M PA R I N G PA R T Y S Y S T E M S
from office and are replaced at time t + 1 by a new government made up of a party or coalition of parties which were previously in opposition. The British case offers the most obvious example of such wholesale alternation, with a single-party Labour government being replaced by a single-party Conservative government, or vice versa. A similar pattern was long evident in New Zealand, with the alternation between Labour and the National party. But while the classic two-party model offers the most obvious examples of wholesale alternation, the pattern can also be seen in more fragmented systems. In Norway, for example, wholesale alternation has regularly ensued on the basis of shifts between a single-party Labour government, on the one hand, and a multiparty bourgeois coalition, on the other, reflecting a pattern of competition similar to that which developed in Costa Rica in the 1960s and 1970s. More unusually, the recent French experience has witnessed wholesale alternation between two competing coalitions, with the Socialists and various left-wing allies, including the Communist party, alternating with a coalition of the Gaullists (RPR) and center–right (UDF). This pattern was also echoed in Germany in 1998, which experienced its first case of wholesale alternation when a coalition of Social Democrats (SPD) and Greens displaced a long-incumbent coalition of Christian Democrats (CDU–CSU) and Liberals (FDP). These latter cases are quite exceptional, however, in that even in systems with four or more relevant political parties, as was the case over extended periods of time in Ireland, Norway, and Sweden, and as was also the case in Japan in 1993, wholesale alternation has usually involved at least one single-party alternative. The second pattern is partial alternation, in which a newly incumbent government in a parliamentary system includes at least one party which also formed part of the previous government. Germany has provided the most obvious example here, in that all of the governments which held office between 1969 and 1998 included the FDP as a junior coalition partner, with the role of senior partner alternating sporadically between the SPD and the CDU–CSU. The Dutch system also approximated this pattern, with the Christian Democrats (CDA) and, prior to 1977, the Catholic People’s party (KVP), persisting in office through to 1994, albeit with alternating coalition partners.6 Indeed, the major contrast between the traditional German and Dutch patterns of alternation was simply that, in the Dutch case, it tended to be the biggest single party that has remained in government, whereas it was usually the smaller partner in the German case. Similar enduring patterns of partial alternation, albeit without involving such pronounced longterm continuity of one particular partner in office, can be seen in Belgium, Finland, and Luxembourg. However, the most striking example of partial alternation was that provided by the Italian case throughout most of the post-war period, with the Christian Democrats (DC) holding office continually from 1946 to 1994, occasionally as a minority single-party government, and more often as the senior partner in a variable multiparty coalition. The third pattern borders closely on this Italian experience and is marked by a complete absence of alternation, or by non-alternation, in
95
96
C O M PA R I N G D E M O C R A C I E S
2
which the same party or parties remains in exclusive control of government over an extended period of time, being displaced neither wholly nor partially. Switzerland offers the clearest example of non-alternation over time, in that the same four-party coalition has now held office continually since 1959. A similar pattern of non-alternation characterizes what Sartori has defined as predominant-party systems, as in the case of Japan from 1955 to 1993, with the Liberal Democrats holding office almost consistently alone over this 40-year period, in India, where the Congress party held continuous office until its first defeat in 1977, and in Mexico, where the Revolutionary party (PRI) held the dominant position from the 1920s through to the end of the century.
Innovation and familiarity Party systems differ not only in their patterns of alternation, but also in the degree to which the alternative governing formulae are either familiar or innovative—that is, whether or not the party or combination of parties has governed before in that particular format (see also Franklin and Mackie 1983). In the British case, for example, familiarity is everything, and no new governing formula has been experimented with since the broad coalition which held office during the Second World War. The formulae were also familiar and hence very predictable over a long period in the Irish case, with governments being made up of either Fianna Fáil on the one hand, or a coalition of all the remaining parties on the other, as well as in Germany, which experimented with no new formula between the advent to office for the first time of an SPD–FDP coalition in 1969 and the eventual incorporation of the Greens for the first time in 1998. Notwithstanding the German case, however, it is in such systems of partial alternation that the greatest scope exists for innovation. In Italy, for example, despite the long-term dominance of the DC, there has been frequent experimentation with new coalition alliances. In the Netherlands, despite the continuity in office of the KVP and, later, the CDA, innovation has also been particularly marked, with differing and novel combinations of parties succeeding one another in office with remarkable frequency. It is important to note here that while innovative formulae are obviously involved when the party or parties concerned have never previously held office, they can also be seen to occur even when the parties have governed before but never in that particular alliance. In the Irish case, for example, the first ever coalition between Fianna Fáil and Labour, which took office in 1993, can be defined as innovative, even though each of the parties involved had already had long experiences of government; similarly, both the People’s party (ÖVP) and Social Democrats (SPÖ) single-party governments in Austria (taking office for the first time in 1966 and 1970, respectively) can be treated as innovative, despite the fact that both parties had previously governed together in coalition.
C O M PA R I N G PA R T Y S Y S T E M S
Which parties govern? The third factor involved here concerns the range of parties which gain access to government. Although not all possible parties can be expected to win a share of office, even with frequent alternation, party systems can nevertheless be distinguished in terms of the degree to which access to office is widely or narrowly dispersed. In other words, party systems can be distinguished in terms of whether all relevant parties eventually cross the threshold of government, as is more or less the case in the Netherlands, for example, or whether governing remains the privilege of just a limited subset of parties, as was the case in post-war Italy. Knowing the range of parties with access to office therefore allows us to distinguish between these latter cases, which otherwise tend to coincide in terms of both of the other criteria indicated above. That is, while both the Netherlands and Italy are similar in the sense of their pattern of partial alternation, in terms of the longevity in office of a core centre party, and in terms of their resort to innovative formulae, they are nevertheless strikingly dissimilar when it comes to the range of parties gaining access to office, with particular parties being persistently excluded in the Italian case prior to the mid-1990s, and with virtually no substantial parties being excluded in the Dutch case. This distinction may also be related to what Sartori defines as polarized pluralism, which, as noted above, required the presence of anti-system parties at each end of the political spectrum, such parties being defined in part as those which were out of competition for government, thus forcing governments to be formed across the span of the centre. Here, also, the concern is with whether certain parties are persistently excluded as unacceptable partners in office. Where this criterion differs from that used in the definition of polarized pluralism, however, is that the question of whether or not such parties are genuinely and objectively “anti-system”, which has always been a point of dispute in the interpretations and criticisms of Sartori’s model, becomes irrelevant (see the comprehensive discussion of varieties of anti-systemness in Capoccia 2000). Rather, what matters is whether there are parties which are treated, in practice, as “outsiders”, and which are regarded by the other parties in the system as unacceptable allies. In this sense, anti-systemness, like beauty, here lies in the eyes of the beholder. It is difficult, for example, to determine whether the right-wing Danish Progress party is genuinely anti-system; on the other hand, it is relatively easy to see that this party has been regarded by its potential allies as an outsider and, up to now, has always languished in opposition (unlike, say, the more ostensibly “anti-system” Communist party in France). Conversely, while the Austrian Freedom party (FPÖ) under the leadership of Jörg Haider has often been regarded as an anti-system party of the extreme right, it has nevertheless been recently incorporated as the junior partner in an innovative coalition government with the ÖVP. In practice, at least, the party is now ‘within’ the system.
97
98
C O M PA R I N G D E M O C R A C I E S
2
Structures of party competition: closed or open? The combination of these three criteria yields a fairly broad-brush distinction between two contrasting patterns in the structure of party competition (see Table 4.2). On the one hand, the structure of party competition can be relatively closed, and hence highly predictable, with little or no change over time in the range of governing alternatives or in the pattern of alternation, and with new parties and/or “outsider” parties finding it virtually impossible to break through the threshold of government. The British case has afforded perhaps the most obvious example to date of such a closed system, being persistently characterized by wholesale alternation, by a complete absence of innovative formulae, and by the presence of just two governing, and governable, parties. The presidential party system in the United States is similarly closed. On the other hand, the structure of party competition can prove relatively open, and hence quite unpredictable, with differing patterns of alternation, with frequent shifts in the make-up of the governing alternatives, and with new parties gaining relatively easy access to office. The post-war Dutch pattern comes quite close to this form, in that new parties have been relatively easily incorporated into government (such as Democratic Socialists 70 in 1971 and both Democrats 66 and the Radicals (PPR) in 1973) and in that innovative formulae have been adopted in almost half of the new governments formed since 1951. Where the Dutch system deviates from a wholly open pattern, however, is in the long-term presence in government of the KVP and, later, the CDA, and in the fact that alternation has always been partial. In this sense, and at least prior to 1994 when the first ever “secular” government was formed, there was always a certain element of predictability involved, and to this extent the structure of competition was at least to some extent closed. Denmark in the post-war period also comes quite close to an open pattern, having experiences of both partial and wholesale alternation, having frequently adopted innovative formulae (in the case of almost one-third of all post-war governments), and having also proved to be relatively open to new parties, such as in 1982, when the Centre Democrats and the Christian People’s party were first admitted to government. On the other hand, even Denmark can be seen as partially closed and somewhat predictable as a result of the persistent exclusion from office of the Progress party and the Socialist People’s party. These examples also underline the extent to which the development of a closed structure of competition owes much to the strategies of the established parties, and, in particular, their unwillingness to experiment with innovative formulae and their reluctance to admit new parties into government. In some instances, of course, the parties may feel themselves genuinely constrained in this respect, in that any new governing options might require the bridging of what are believed to be ineluctable divides in policy and/or ideology. The structure of competition may therefore be very predictable, and hence closed, as a result of the distances which separate the
C O M PA R I N G PA R T Y S Y S T E M S
TABLE 4.2 Closed and Open Structures of Competition Closed structure of competition Open structure of competition Wholesale alternation in office, or non-alternation Partial alternation, or mix of both partial in office and wholesale alternation Familiar governing formulae Innovative governing formulae Access to government restricted to a limited Access to government open to (almost) number of parties all parties Examples: United Kingdom, New Zealand (to mid-1990s), Japan (1955–93), Switzerland, Ireland (1948–89)
Examples: Denmark, the Netherlands, post-authoritarian party systems
relevant parties along any one of a variety of different dimensions of competition. Such arguments have been used by DC leaders in Italy, for example, in order to justify the persistent exclusion of the PCI from office, and are equally cited by a number of different party leaderships in Denmark in order to justify the persistent exclusion of the Progress party. In other instances, however, it is obvious that the maintenance of familiar and closed patterns of competition simply constitutes a strategy of self-preservation on the part of the established parties.7 In Ireland, for example, the long-term refusal of the dominant Fianna Fáil party to even consider entering a coalition, a refusal which contributed substantially to the long-term closure of competition in Ireland, was clearly designed to maintain its status as the only party capable of offering single-party government, and was thus intended to maintain its electoral credibility. A similar sense of self-preservation can be seen to characterize the long-term reluctance of the two major British parties to consider the possibilities of coalition with the smaller Liberal party, even though Labour did come strikingly close to such a path-breaking option during the Lib–Lab Pact in the late 1970s, and although the Liberals were invited to join certain government committees following New Labour’s victory in 1997. To be sure, there are real limits on the capacity, and willingness, of the established parties to maintain a closed structure of competition. New parties might emerge which have to be taken on board; particular party leaders may have their own agendas and priorities; external crises might develop which force the adoption of new strategies, and so on. Nonetheless, any explanation of the degree of closure of any given structure of competition must necessarily focus particular attention on the strategies of the parties themselves. Closed structures of competition are clearly characteristic of traditional two-party systems, and, of course, of those systems which have experienced a real absence of alternation over time, such as Japan before 1993, the former Stormont regime in Northern Ireland, Mexico before 2000, Singapore, or Switzerland. Conversely, openness, and a lack of predictability, tends to characterize more fragmented systems (that is, systems with a relatively large number of relevant parties), which experience partial alternation and which often lack one large core party of government. Moreover, since closure
99
100
C O M PA R I N G D E M O C R A C I E S
2
necessarily requires the development of stable norms and conventions in the patterns of competition and in the processes of government formation, it is also clearly a function of time, and, most crucially, is not something which can be seen to characterize the party “systems” that emerge more or less from scratch following post-authoritarian democratization. Indeed, what is most striking about such new party systems, including those which have emerged in post-communist Europe, for example, as well as those of Spain and Portugal through to the 1980s, is precisely their lack of closure and hence their lack of systemness. Seen from this perspective, the long-term process by which party systems may eventually become consolidated can also be seen as a long-term process by which the structure of competition becomes increasingly closed and predictable. Thus, while a more closed and predictable structure might well develop in a number of the post-communist democracies in the next decade or so, this can prove a relatively lengthy process (but see Toole 2000). The fruits of such a longterm process of structural consolidation can now be seen in the relatively recently democratized Portuguese and Spanish systems (e.g., Morlino 1998), and may also be already beginning in the Czech Republic, not least as a result of the efforts of the two main parties to modify the Czech institutions in a way that is likely to penalize their smaller rivals. Long-term structural consolidation can also be seen in some of the Latin American systems, with one of the most notable examples being the strengthening of two-partyism in Costa Rica. In this latter region, however, as in the United States, the frequent combination of a presidential system of government, on the one hand, and the presence of often undisciplined parties (e.g., Mainwaring and Scully 1995a), on the other, suggests that however closed presidential party politics might be, there remains a significant bias against the development of closed structures of competition in the legislative electoral arena.8 A similar process can now be identified in Israel (e.g., Medding 2000), where the recent institution of a separate election for the position of the prime minister has effectively undermined the need for parliamentary discipline, and has led to increased fragmentation in the legislative arena and to a wholesale de-structuring of the traditional party system. In short, the degree of closure varies, ranging along a continuum from situations in which it is least pronounced, such as in post-authoritarian party systems, to those in which it is most pronounced, such as in those established systems in which there is little or no innovation in the processes of government formation, and in which new parties rarely if ever break through the governing threshold.9 A similar pattern of variance is highlighted in Mainwaring and Scully’s (1995b) overview of Latin American systems, where they distinguish between “institutionalized”—or what I might call “closed”—party systems, on the one hand, and “inchoate”—or “open”—party systems, on the other. The real question here, however, is whether we can conceive of open or inchoate “systems” as being genuine systems at all. The degree of “systemness” certainly varies from one party system to another. But it is only in
C O M PA R I N G PA R T Y S Y S T E M S
cases where this degree of systemness becomes pronounced, and where the party system approaches closure or institutionalization, that we can properly speak of there being a party system as such. The idea of a wholly open or wholly inchoate party system may therefore amount to a contradiction in terms. In a similar vein, it can also be argued that the well known Lipset and Rokkan (1967) formulation regarding the “freezing” of party systems can be read as simply another way of referring to the progressive institutionalization or closure of these systems, and hence to speak of the freezing of party systems may simply be another way of saying that collections of parties eventually stabilize their patterns of interactions and thus develop into systems. As systems, they are by definition frozen and institutionalized; de-freezing, in this reading, is system failure (see also Mair 2001).
Party systems and electoral outcomes The notion of the closure or openness of party systems’ structure of competition is also important in that it immediately allows us to move away from the conventional idea that party system change is largely, if not exclusively, a function of, or even a synonym for, electoral change. In other words, it affords a conception of party system change which may owe its origin to factors other than simply the flux in voter preferences. For although party system stability and change, on the one hand, and electoral stability and change, on the other, may certainly be related, they are far from being synonymous. Electoral alignments might shift, for example, even in quite a dramatic way, without necessarily impinging significantly on the structure of competition, and hence without necessarily altering the character of the party system itself. Conversely, the structure of party competition and hence the nature of the party system itself might suddenly be transformed, even without any significant prior electoral flux. One telling, if now quite dated, example comes from Denmark, where the 1973 election witnessed one of the most substantial electoral shifts to have occurred in post-war Europe up to that time, resulting in an immediate doubling of the number of parties represented in parliament (see Pedersen 1988). Prior to 1973, five parties had been represented in the Danish Folketing, together accounting for some 93% of the total vote. As a result of the 1973 election, five new parties won representation, and the total vote won by the previously represented parties fell to less than 65%. This was a massive shift by any standards, and since the new entrants to the parliament included both the long-established Communist party as well as the newly formed right-wing Progress party, it also resulted in a major increase in the level of polarization. In practice, however, it is certainly possible to question whether this change had any substantial impact on the workings of the Danish party system. To be sure, a new government had to be formed, which, in fact, was a minority, single-party Liberal government, the first such to take office since 1945. On the other hand, this innovative
101
102
C O M PA R I N G D E M O C R A C I E S
2
government was then succeeded by a Social Democratic minority government, which was precisely the same formula that had been employed prior to the 1973 earthquake election, and then eventually by a center–right coalition, which was differently composed but otherwise essentially similar to the various other center–right coalitions which had governed Denmark in the early 1950s and late 1960s. The increased fragmentation and the greater degree of polarization after 1973 did of course make governing more difficult: it was not until early 1993 that a government was to enjoy majority status in the parliament (although such a status had also been quite exceptional even prior to 1973); post-1973 governments also tended to collapse more frequently than before; there was a more frequent resort to elections; and finally, as noted above, new parties had eventually to be accommodated into government, although the Progress party has not yet been accorded this particular privilege. But the question still remains as to whether this relatively massive shift in electoral preferences had any real effect whatsoever on the structure of party competition and therefore on the party system itself. Denmark is now, but always has been, quite innovative in terms of governing formulae; it is now, and always has been, reasonably open to new parties coming into government; and now, as before, it experiences both wholesale and partial alternation in government on a regular basis. Moreover, it had once, and still maintains, a relatively open structure of competition and hence a relatively unconstraining party system. In these terms at least, it appears that 1973 has not made any significant difference. The Italian case in the 1990s also offers a useful example, even though uncertainty still remains as to the direction, if any, in which the party system is being transformed (see especially Bartolini and d’Alimonte 1998). On the face of it, there is no other established Western party system which has undergone such a profound change. At the electoral level, for example, following decades of relative stability, the 1994 election resulted in a level of volatility of more than 37%, which is not only the highest figure recorded in Italian history, but, even more strikingly, is substantially higher than that recorded in almost any election held in Western Europe between 1885 and 1989.10 In terms of format, the system has also became totally transformed, with the emergence of new parties and the reconstitution of established parties leading to a situation in which not one of the parties represented in parliament in the late 1990s had also been represented under the same name or in the same form as recently as the late 1980s. Finally, it can also be argued that there has also been a major change in the level of polarization, as a result of the transformation of the mainstream of the Communist party (PCI) into the Left Democrats (DS), on the one hand, and its eventual leadership of government in 1996, and, on the other, the transformation of the neo-fascist Social Movement (MSI) into the National Alliance (AN) and its incorporation into government in 1994. These are certainly profound changes. The most relevant question, however, is whether these changes will have any long-term impact on the structure of competition. The structure of competition in the “old” party
C O M PA R I N G PA R T Y S Y S T E M S
system was certainly clear. Governments were formed from out of the center, were dominated by the Christian Democrats, and involved shifting alliances with partial alternation across the center–left and center–right while excluding both the PCI and the smaller MSI. Any fundamentally new pattern might therefore seem to require that governments be formed almost exclusively from the left or from the right (creating the potential for wholesale alternation), that the extremes disappear or be incorporated, and, in this new bipolar world, that the independent position of the center be marginalized. With the formation of the right-wing Berlusconi government in 1994, it certainly seemed possible that this was the pattern that was beginning to emerge, and this seemed to be confirmed when that innovative government was then displaced by an alternative but equally innovative coalition of the center–left, dominated by the DS, which, in turn, was displaced by a reconstituted Berlusconi coalition in 2001. There remains a question as to what will happen to any new center, however, in that a number of centrist parliamentary groups have now been formed which seek to play off left against right, and which appear to be trying to establish an independent pivotal position in the system. As of now, of course, it is too soon to suggest what future patterns or alignments might yet emerge, and in this sense the structure of competition might be regarded as quite open and unpredictable, which, at least in the short term, does represent a fundamental change. Indeed, in this sense the “system” looks increasingly unstructured and hence transformed. Should the center finally split into its own “left” and “right,” then this would cement the emerging bipolar pattern, and this would certainly lead to a new structure of competition and hence to a wholly new party system. But should an independent center manage to reconstitute itself, and should it prove capable of taking full advantage of its pivotal position by playing left off against right in a way which would allow it to construct a broad alliance across the center, then, despite the different actors and their different weights in the system, we might well end up by witnessing the recreation of more or less the same structure of competition as had prevailed prior to the 1990s (Bartolini and D’Alimonte 1998). Question marks also hang over the real extent of change in the Canadian party system, notwithstanding the electoral earthquake of 1993. In this case, as in Italy, the level of volatility rose to an unprecedented and massive high— 42%—almost five times that of the average level of volatility recorded in the 1970s and 1980s. The consequences were also very far-reaching, with the once powerful Progressive–Conservative party being reduced to just 16% of the vote in 1993 (its lowest share since 1949), and, even more strikingly, to just two seats (as against 169 in the previous election). In addition, two new parties, the Reform party and Bloc Québécois, won substantial representation in parliament, the first parties outside the mainstream to do so since the effective demise of Social Credit in the mid-1970s. Moreover, precisely because these two new parties were so evidently regional in character, they also signalled a potentially enduring shift in electoral alignments. This was substantial change by any standard, and some Canadian scholars have argued that it
103
104
C O M PA R I N G D E M O C R A C I E S
2
represents the dawn of a new and distinctively different party system which has yet to fully develop (Carty, Cross, and Young 2000). At the same time, however, there was also one striking continuity, in that the Liberal party returned to government with a powerful majority in 1993, retaining that position with decisive election victories in 1997 and 2000. This suggests the possible return to a “one party dominant” system, similar to that which characterized Canadian party politics in the 1940s and 1950s. At one level, the system has certainly changed: the Liberals now confront an opposition divided across four parties instead of just two, and serious regional divisions have erupted onto the federal stage (Nevitte et al. 1999). An attempt to restructure the Reform party into a new, more broadly based, opposition party (the Canadian Alliance) prior to the 2000 election failed to improve significantly on the electoral gains achieved by Reform when it became the official parliamentary opposition to the Liberals in 1997. While this process of attempting to construct a viable conservative alternative continues, it is as yet impossible to speak of any enduring transformation in the structure of competition in federal politics under the present conditions of uncertainty regarding the future of the Alliance. Meanwhile, the Bloc Québécois continues to provide the only significant opposition at the national level to the Liberals in Quebec, building on the solid base established there in provincial politics by the Parti Québécois since the 1970s. Finally, we can turn to the Japanese case. Here, in brief, we can also see signs of an Italian-style meltdown, with the traditionally predominant Liberal Democrats (LDP) finally losing office in 1993, following four decades of unbroken rule, in the wake of a highly volatile election in which three new parties suddenly emerged to win some 20% of the seats in the House of Representatives. Three years later, and as the result of further splits, mergers, and formations in what had become a very unstable party environment, two other new groups emerged, the New Frontier party (NFP) and the Democratic party (DPJ), together winning some 40% of the seats. Moreover, and as in Italy, the meltdown of the old order was associated with the formation of innovative coalitions and the arrival in office of parties which had previously either languished in opposition or had just recently been formed for the first time (Mair and Sakano 1998; Watanuki 2001). In effect, then, the old structure of the LDP predominant-party system had dramatically collapsed. In contrast to Italy, however, where no new structure of competition has yet been consolidated, and hence where the party ‘system’ remains quite inchoate, the Japanese case seems to have adjusted quite quickly to a new form of competition, in which the now diminished but still powerful LDP has taken on the role of a major center party, building a variety of—sometimes short-lived—coalitions to its left and to its right, and thereby returning to office as the power-broker par excellence. Thus although the continued survival of the LDP has meant that the changes to the party format have not proved quite so dramatic as in Italy, the new balance of forces has nevertheless led to the emergence of a wholly novel structure of competition.
C O M PA R I N G PA R T Y S Y S T E M S
What we see here, then, is one instance in which substantial electoral change does not appear to have led to significant party system change (Denmark); one instance in which electoral change may be associated with a major shift in the structure of competition (Japan); and two instances in which, despite extraordinary electoral flux, question marks still remain as to whether a new type of party system might yet develop (Canada and Italy). It is precisely the combination of these different experiences which underlines the need to separate out the notion of party system stability/change, on the one hand, and electoral stability/change, on the other. Not only that, however, for what may be most interesting about the separation of these two processes, and about the recognition that change in party systems may be due to factors other than electoral change, is that it also affords the opportunity to reverse the conventional chain of influence, and to probe the extent to which party system stability (or change) may lead to electoral stability (or change), rather than simply the other way around. Electoral alignments are, of course, stabilized by a variety of factors, including, most crucially, the cleavage structure (Lipset and Rokkan 1967). Other factors that play a role here include the constraints imposed by institutional structures, such as by the electoral system, as well as those deriving from the organizational efforts of the parties themselves (Bartolini and Mair 1990). In the case of Brazil, for example, as Mainwaring (1998) argues, much of the exceptionally high electoral volatility can be explained not just by social structure, by also the particular character of Brazilian political institutions and by the behavior of the party elites. Even within this broader perspective, however, one additional “freezing” agent that is often neglected is the effect of the structure of party competition itself. As should be evident, a closed structure of competition clearly constrains voter preferences, in that it limits the choice of governing options in a way which is similar to the limits on the choice of parties in non-fragmented systems. A closed structure of competition therefore also clearly enhances party system stability, and, indeed, helps to ensure that party systems generate their own momentum and thus freeze into place. In short, the stabilization of electoral alignments has at least partly to do with the consolidation of a given structure of competition. What this also implies, of course, and perhaps most interestingly, is that a change in that structure may then itself serve to destabilize established electoral alignments. In Italy, for example, the basis for a wholesale change in electoral preferences in 1992 and 1994 was at least partially laid by the “legitimation” of the Democratic Party of the Left, which undermined the terms of reference by which Italian party competition had been structured since the late 1940s. Italian voters, as well as the Italian parties themselves, had long been constrained by the belief that there was no alternative to a Christian Democratic-dominated government. And once such an alternative finally did emerge through the transformation of the unacceptable Communist party into the highly acceptable Democratic Left, this particular anchor was cut loose, and voters began to shift in relatively great numbers,
105
106
C O M PA R I N G D E M O C R A C I E S
2
eventually leading to the virtual disappearance of the once dominant Christian Democratic party. In yet another case, in Ireland, and following decades in which there had been no major changes in the electoral balance of the party system, the long-term basis for stability was finally undermined in 1989 when the dominant party, Fianna Fáil decided for the first time ever to enter a coalition with another party. Prior to then, party competition had been structured around the opposition between Fianna Fáil on the one side, and all of the smaller parties on the other, and this had severely constrained and stabilized voter preferences. From 1989 onwards, however, when these constraints were removed, the potential for change was greatly enhanced, and hence while the degree of electoral instability prior to Fianna Fáil’s first coalition was relatively muted, the subsequent election witnessed a major upsurge in volatility which resulted in the traditionally marginal Labour party doubling its vote—the result, quite simply, of the removal of what had been up to then the most powerful constraint on electoral mobility. This is perhaps a roundabout way of saying that the structure of competition, and the structure of competition for government in particular, may impose a major constraint on voter choice, and hence may act to stabilize electoral alignments. In this sense, voters are not simply expressing preferences for individual parties; rather, albeit not always to the same degree in different party systems, and this in itself is an important source of crossnational (and cross-institutional) variance, they are also expressing preferences for potential governments. And in much the same way that a shift in the range of parties on offer can act to undermine established preferences, so too can a shift in the range of governing options, and hence a shift in the structure of competition, act to undermine established preferences and promote electoral instability. In short, it is not only a question of electoral change leading to party system change, as does seem to have been the case in Japan; rather, it can also be the other way around. Party systems do not simply reflect electoral preferences. They also serve to constrain them.
NOTES
1. See, for example, Almond et al. (1993: 117–20), where this traditional distinction is recast as one of “majoritarian” versus multiparty systems; see also the influential study by Lijphart (1999), where one of the key distinctions between majoritarian and consensus democracies is defined as that between a two-party system and a multiparty system, although in contrast to the traditional approach, in Lijphart’s case it is the multiparty variant that is associated with consensus politics. 2. For earlier versions of this typology, see Sartori (1966, 1970). An edited version of the Sartori typology, together with those of the other authors discussed here, is reprinted in Mair (1990). 3. Although the predominant-party system constitutes a useful category, it fits rather uneasily into Sartori’s framework, since it is defined by wholly different criteria, and can by definition co-exist with every possible category of party
C O M PA R I N G PA R T Y S Y S T E M S
4. 5. 6. 7.
8.
9. 10.
numbers (that is, it can develop within a context of a two-party system, a system of limited pluralism, and a system of extreme pluralism) and, at least theoretically, with every possible spread of ideological opinion. The exception would be those systems which might be categorized as predominant-party systems, and which, as noted above, do not easily fit into the criteria adopted for the typology as a whole. For an application of these criteria to existing party systems in Europe, see Luther and Deschouwer (1999) and Toole (2000). Following the 1994 election, and for the first time since the advent of full democracy, a government was formed in the Netherlands without the inclusion of the Christian Democratic/religious mainstream. It is in this sense that such closure also involves what Schattschneider (1960: 69 and passim) has defined as “the mobilization of bias,” with the emphasis on particular conflicts and on distinct alternatives acting to preserve the interests of the various protagonists that align themselves in terms of these conflicts. The lack of structuring is also reflected in the persistently high levels of electoral volatility which are strikingly and substantially higher than those occurring in the relatively structured west European systems (see Coppedge 1992; Mainwaring 1998: 535–7). As noted, closure is also probably least pronounced in the legislative electoral arena in presidential systems, notwithstanding any strong structuring which might be evident in the presidential electoral arena in these same systems. The index of volatility measures the net aggregate shift in votes from one election to the next, and is the equivalent of the total aggregate gains of all winning parties or the total aggregate losses of all losing parties (see Pedersen 1979). The average volatility in post-war Europe has been less than 9%, and, apart from the 1994 election in Italy, only four other European elections in the past century have exceeded 35%: Germany in 1919 (47.5%), France in 1945 (36.4%), and Greece in 1950 (47.0%) and 1951 (45.1%)—see Bartolini and D’Alimonte (1995: 443–4); on electoral stability and instability in Europe more generally, see Bartolini and Mair (1990).
107
